ADAPTIVE SUB-BAND ACOUSTIC ECHO CANCELLATION ALGORITHMS FOR ECHO PATHS INVOLVING NONLINEAR AMPLIFIERS AND TRANSDUCERS

摘要

This paper examines the normalized least-mean-squares (NLMS) algorithm, the proportionate normalized least-mean-squares (PNLMS) algorithm, and the affine projection algorithm (APA) applied to the linear and nonlinear sub-band acoustic echo canceller (AEC). The AEC aims at reducing the acoustic echo generated from a loudspeaker-enclosure-microphone (LEM) system. In order to enhance the efficacy of echo cancellation, the AEC utilizes sub-band structure in conjunction with the PNLMS algorithm and the APA. To cope with nonlinear distortions from low-cost transducers, the AEC is augmented with nonlinear processors based on second-order Volterra filters and a Hammerstein nonlinear model. The proposed AEC was verified through extensive simulations and experiments. The results have shown that the proposed system achieved improved echo cancellation performance over conventional systems and met the telecom standard ITU-T Rec. G.168.